The Murine MHC Class II Super Enhancer IA/IE-SE Contains a Functionally Redundant CTCF-Binding Component and a Novel Element Critical for Maximal Expression Free

Located on chromosome 17, the murine MHC class II (MHC-II) region spans 250 kb (1, 2) and encodes, in many murine haplotypes, I-A and I-E α/β heterodimeric MHC-II molecules. MHC-II molecules are expressed on the surfaces of Ag-presenting cells and function to present antigenic peptides to CD4 T lymphocytes for initiation and/or regulation of adaptive immune responses. However, because of a promoter region and first exon deletion, some haplotypes, such as the H-2^b haplotype of the C57BL/6 mouse, do not express the I-Eα–chain gene H2-Ea (3), resulting in the expression of only one functional MHC-II molecule, I-A. In addition to the classical MHC-II genes, other genes associated with Ag processing, such as those encoding H2-DM and H2-O, as well as some genes that function in MHC-I Ag presentation (Tap1, Tap2), are located within this locus (4).

Expression of MHC-II genes is cell type specific and regulated at the level of transcription (reviewed in Ref. 5). Within a few hundred bases from the transcriptional start sites of each MHC-II gene are conserved regions (W, X1, X2, and Y boxes) that are essential for MHC-II transcription (6, 7). Regulatory factor X (RFX) (8), CREB (9), and NFY (7) bind constitutively and cooperatively to the X and Y boxes (10), respectively. Alone, these DNA binding factors are not sufficient to activate MHC-II gene expression, requiring the transcriptional coactivator termed CIITA (11). CIITA expression itself is tightly regulated and is the limiting factor governing MHC-II gene expression. When expressed, CIITA is recruited to the RFX–CREB–NFY complex and mediates additional interactions between chromatin remodeling complexes and various components of the general transcription machinery (12–15).

The mammalian insulator factor, CCCTC binding factor (CTCF), functions to organize and regulate higher order or three-dimensional chromatin structure in eukaryotes by binding to its cis-acting elements and connecting two distantly separated DNA fragments (16–18). This is accomplished by forming CTCF dimers and bridging distant CTCF binding sites, leading into chromatin loops. CTCF elements also function as boundaries between active and inactive heterochromatin domains (19). Such boundaries could function to block the activity of enhancer elements from acting on nearby promoters and can insulate genes from heterochromatic silencing (19). The formation of local CTCF-dependent chromatin loops was particularly striking in the human MHC-II locus, where CTCF binding sites occurred at the boundaries between subregions encoding MHC-II α/β gene pairs (e.g., HLA-DR and HLA-DQ) and chromatin loops/interactions occurred between adjacent regions (20, 21). These CTCF binding site interactions were independent of the above MHC-II–specific regulatory factors, including CIITA. RNA interference–mediated knockdown of CTCF in the human Raji B lymphoblastoid cell line resulted in decreased MHC-II gene expression and loss of the interactions between the CTCF sites, suggesting that these interactions were important for maximal MHC-II expression (21). Additional interactions between the CTCF sites and CIITA-bound MHC-II promoters were also observed, suggesting that a complicated set of interactions were critical for maximal MHC-II expression (22, 23).

In murine B cells, a series of interacting CTCF sites were identified within the murine MHC-II locus that appeared to play a similar role to the human system (24). Intriguingly, these CTCF–CTCF interactions were completely reorganized as B cells differentiated to plasma cells, suggesting that expression of the MHC-II genes was dependent on this architecture (24). Plasma cell differentiation leads to the loss of both CIITA and MHC-II gene expression (25–27). Stable, ectopic expression of CIITA in a plasma cell line was used to reactivate MHC-II expression and potentially reprogram the architecture of the locus. Despite the ability to generate high levels of MHC-II expression and induce interactions between MHC-II promoters and some CTCF sites, the three-dimensional architecture of the locus remained mostly in the plasma cell configuration. Thus, other factors/elements or processes might regulate the chromatin architecture of the region. Recently, in human B cells, a super enhancer (SE) was identified and termed the DR/DQ-SE as it was located between these two human MHC-II gene pairs (28). Deletion of the DR/DQ-SE altered MHC-II expression and the local chromatin architecture, defining a new regulatory mechanism controlling MHC-II expression (28).

In this study, similar to the DR/DQ-SE in humans, the entire region between the I-A and I-E genes was identified as an SE in the murine MHC-II locus and termed IA/IE-SE. Within the intergenic region are two major components: a strong and broad peak of histone H3K27 acetylation (ac) termed region b and a CTCF binding site [previously termed C15 (24)]. C15 has moderate CTCF-binding activity that is lost upon plasma cell differentiation (24). A knockout mouse that deleted C15 was made to understand its role in MHC-II expression. We found that C15 was functionally redundant with other CTCF sites as I-A expression was maintained in all immunological cell types examined, and chromatin architectural interactions were repositioned to other CTCF sites. Surprisingly, although IFN-γ induction of MHC-II expression in macrophages of C15-deficient mice was not affected, it was abolished in primary embryonic fibroblasts derived from these mice, suggesting that C15 may play a role in distinct cell lineages. CRISPR-mediated deletion of a PU.1 binding site within region b resulted in reduced H2-Aa and H2-Eb1 expression and alterations in chromatin looping, demonstrating an important role for this region in MHC-II expression and identifying a novel and direct role for PU.1 binding in regulating MHC-II expression. Thus, the results presented in this study demonstrate flexibility in the use of CTCF sites and add a new role for PU.1 and an SE in regulating the expression of MHC-II genes.

The pM30 plasmid (29) used in this study was previously modified to contain the SalI and MluI sites and a second LoxPsite (29). Homology targeting fragments surrounding C15 (genome build mm10, chr17:34296672-34299610 [5′] and chr17: 34300305-34303340 [3′]) were inserted into the SacII site and MluI–KpnI sites of the pM30 plasmid (29), respectively (Supplemental Fig. 1). The C15 fragment (chr17:34299611-34300304) was inserted between two LoxP sites. The resulting targeting construct contained the 5′ homology arm, two LoxP sites surrounding the C15 fragment and Frt-flanked neo cassette, and the 3′ homology arm. All inserted fragments were verified by PCR, as well as sequencing. The Mouse Transgenic and Gene Targeting Core at Emory University transfected C57BL/6 ES cells with the targeting construct. The ES cells were screened by Southern blot analysis, and the positive ES clones were used to produce C15^fl/fl mice. Founding mice were bred to B6.FVB/N(Ella-Cre)C5379Lmgd/J (stock no. 003724, The Jackson Laboratory) to delete C15 sequences. Sequences for all PCR primers used to perform genotyping are listed in Supplemental Table I. Through inbreeding, lines that contained homozygous deletions of C15 and were Cre⁻ were expanded for further analyses and are referred to as ΔC15 mice.

C57BL/6 mice were purchased from The Jackson Laboratory and bred in-house. Mice were housed in the Emory University School of Medicine Facilities. All animal experiments were approved by the Emory University Institutional Animal Care and Use Committee. Wild-type (WT) C57BL/6 and ΔC15 mice between 6 and 8 wk of age were used for all experiments.

Splenic CD43⁻ B cells were isolated using the Miltenyi B Cell Isolation Kit (Miltenyi Biotec; 130-090-86) and checked for purity by flow cytometry using Abs for CD43 and B220. For B cell differentiation experiments, cells were cultured at a concentration of 0.5 × 106 cells/ml in B cell media (RPMI 1640 supplemented with 10% heat-inactivated FBS, 1× nonessential amino acids, 0.05 mM 2-BME, 1× penicillin/streptomycin, 10 mM HEPES, and 1 mM sodium pyruvate) containing 20 mg/ml Escherichia coli O111:B4–derived LPS (Sigma-Aldrich; L2630), 5 ng/ml IL-5 (eBioscience; 14-8051), and 20 ng/ml IL-2 (eBioscience; 14-8021) as previously described (27). On day 3 of culture, a subset of cultured cells was analyzed for differentiation and MHC-II surface expression by flow cytometry (LSRFortessa, BD Biosciences), whereas the remaining cells were used to sort 1 × 10⁶ CD138⁺ cells. Purified naive B cells and sorted CD138⁺ cells were resuspended in 600 μl of RLT buffer (QIAGEN; 79216) containing 1% 2-ME and snap frozen for RNA isolation.

Mouse embryonic fibroblasts (MEF) were isolated from C57BL/6 and ΔC15 mice using the standard protocol (30). Mice were set up for breeding and checked for evidence of mating the next day. Embryos were removed on embryonic day 13.5, washed with PBS, and transferred to a tissue culture dish containing 5 ml trypsin/EDTA solution (Corning). Embryos were dissociated by passing through a 16-gauge needle three times. An additional 5 ml of the trypsin/EDTA solution was added to the dish, and the contents were chopped with a sterile razor blade before incubation at 37°C for 10 min. The samples were chopped again before an additional 5-min incubation. The contents were transferred to a conical tube with an equal volume of MEF medium (DMEM with high glucose [catalog (Cat) no. D5796, Sigma], sodium pyruvate, 5% heat-inactivated FBS [HyClone Laboratories], 5% heat-inactivated bovine calf serum [HyClone Laboratories], 1× nonessential amino acid, 2 mM l-glutamine, 0.1 mM 2-ME, 20 mM HEPES [pH 7.3], and 1× penicillin/streptomycin) and left standing for 3 min. The medium above the settled tissues was centrifuged for 5 min at 1000 × g, and the cell pellet was resuspended in MEF medium and plated into a 100-mm culture dish.

Primary bone marrow–derived macrophages were prepared from the femurs of C57BL/6 and ΔC15 mice as previously described (31). Bone marrow cells were cultured for 7 d in DMEM with 10% FBS, 100 U/ml penicillin/streptomycin, and 20% L-929 cell line–conditioned media, which contains cell line–produced M-CSF. After 7 d in culture, nonadherent cells were eliminated, and adherent cells were harvested for assays. The murine fibroblast cell line L-929 was obtained from American Type Culture Collection (CCL-1) and cultured in DMEM supplemented with 10% FBS and 100 U/ml penicillin/streptomycin.

The cell lines A20 (TIB-208) and P3X63Ag8 (CRL-1580; termed P3X in this study), representing B cells and plasma cells, respectively, were purchased from the American Type Culture Collection. All murine cells were grown in RPMI 1640 medium (Mediatech) containing 10% heat-inactivated FBS (Sigma), 10 mM HEPES (HyClone Laboratories), 1 mM sodium pyruvate (HyClone Laboratories), 1× nonessential amino acids (HyClone Laboratories), and 0.05 mM 2-ME (Sigma-Aldrich).

For flow cytometry, cells were collected and washed with cold PBS and resuspended in FACS buffer (1 × PBS, 1 mM EDTA, 1% BSA, and 0.2-μm filter sterilized). Prior to Ab staining, cells were incubated with anti-CD16/32 (Fc block clone 2.4G2) on ice for 5 min. To measure the level of cell surface MHC-II expression, cells were labeled on ice with 1:300 dilution of PE-conjugated anti–IA-E Ab (BD Pharmingen), and isotype control (BD Pharmingen) (0.5 ng/ml) was added to all cells. Stained cells were washed twice with FACS sorting buffer. BD LSRII or BD Fortessa flow cytometers were used for data collection, and the FlowJo computer software (Tree Star, Ashland, OR) was used for analyses.

Common lymphoid progenitor (CLP), pre-pro, pro-B, pre-B, immature B, and mature B cells were analyzed from bone marrow cells with a combination of fluorescence Abs as follows: CLP used IgM⁻, CD43⁺, CD19⁻, and B220⁻ cells; pre-B–pro-B cells used IgM⁻, CD43⁺, CD19⁻, and B220⁺; pro-B cells used IgM⁻, CD43⁺, CD19⁺, and B220⁺; pre-B cells used IgM⁻, CD43⁻, CD19⁺, and B220⁺; immature B cells used IgM⁺, CD19⁺, and B220^+(mid); and mature B cells used IgM⁺, CD19⁺, B220^+(hi). Abs for flow cytometry staining were purchased from the following companies: CD19-PerCP-Cy5.5 (BD Pharmingen, Cat no. 551001, clone 1D3), CD43-FITC (BD Pharmingen, Cat no. 553270, clone S7), IgM-BV650 (Invitrogen, MA1194D650, clone PFR-03), B220-Cy7 (BioLegend, Cat no.103222, clone RA3-6B2), and CD11b-allophycocyanin (Tonbo Biosciences, Cat no. 20-0112-U100, clone M1/70). Splenic follicular zone B cells (CD21^+/–CD23^+/–)and marginal zone B cells (CD21^+/–CD23^+/–) were assessed using the following Abs: CD21-allophycocyanin-Cy7 (BD Pharmingen, Cat no. 558658, clone 7G6) and CD23-v450 (Invitrogen, Cat no. 48023282, clone B3B4).

CRISPR/Cas9-mediated deletion methodology was employed to create mutants ΔC15 and ΔN1 in the A20 murine B cell line. Guide RNA sequences targeting regions of interest were selected using the CRISPR design tool (http://crispr.mit.edu/) and listed in Supplemental Table I. Guide sequences were cloned into the pX330-U6-Chimeric_BB-CBh-hSpCas9 plasmid (pX330, no. 42230, Addgene, Cambridge, MA), as described previously (32). The U6 promoter and single guide RNA cassette was created using a two-step PCR. For ΔC15, the U6 promoter containing forward primer (5′-CCTGAGCATTCTCCTTATCC-3′) and the reverse primer (5′-CTAGTTGGTGCGAGAAGGAC-3′) were used with the pX330 plasmid template for 15 cycles of PCR amplification. Similarly, for the ΔN1 targeting, forward (5′-CCACCAGCCATAATGATCAC-3′) and reverse (5′-CTGTCTGTGGGTCTGTACTC-3′) primers were used in a 15-cycle PCR. The resulting PCR product was diluted 1:200 and reamplified for 15 cycles. The single guide RNA PCR products, pX330 plasmid, a 70–90 bp ssDNA oligonucleotide homology arm (in case of ΔN1, CGACAATGACGACGATAACTATGATGTTGTTTGGGGGTGCCTGGTGGGTCCAGCATCATC) (33) and the pFLAG-GFP plasmid (27) to monitor transfection efficiency were cotransfected into the A20 cells using a Nucleofector II (Lonza, Walkersville, MD), according to the manufacturer’s protocol. After 4 d of transfection, single GFP-positive cells were sorted into 96-well plates, and after 3 wk, cells were screened for evidence of deletion by PCR (Supplemental Table I). All homozygous mutants were confirmed by PCR and sequencing.

Total RNA was purified using the RNeasy RNA Isolation Kit (QIAGEN, Valencia, CA). cDNA synthesis was performed using SuperScript II Reverse Transcriptase (Invitrogen, Carlsbad, CA) with 2 µg of total RNA in a volume of 20 µl in PCR II buffer containing 5 mM MgCl₂ (Applied Biosystems, Foster City, CA). After reverse transcription, cDNA was diluted 1:200 µl with double-distilled water (RNase free), and 3 µl was used for quantitative real-time PCR. The levels of 18s rRNA were used in each of the experiments to normalize the input between samples. PCR primers are listed in Supplemental Table I.

Chromatin immunoprecipitation (ChIP) assays were performed as described previously (12). Abs used included anti-H3K27ac (MilliporeSigma, Cat 07-360), anti-H3K4me1 (MilliporeSigma, Cat 07-436), anti-H3K4me3 (MilliporeSigma, Cat 07-473), anti-CIITA (Rockland Immunochemicals, Cat 100-401-249), or anti-IgG (MilliporeSigma, Cat 12-370). Protein G magnetic beads (Invitrogen Cat 2019-12-31) (10 µl/sample) were used to isolate the chromatin–Ab complexes. Precipitated chromatin DNA was purified and quantitated by real-time PCR using a five-point genomic DNA standard curve using a CFX96 Real-Time System C1000 Thermal Cycler (Bio-Rad Laboratories). PCR reactions contained 5% DMSO, 1× SYBR Green (Cambrex, Cat 50513), 0.04% gelatin, 0.3% Tween-20, 50 mM KCl, 20 mM Tris (pH 8.3), 3 mM MgCl₂, 0.2 mM dNTP, and 100 nM of each primer. Sequences for all primers used in the ChIP real-time PCR assays are listed in Supplemental Table I.

A modified chromatin conformation capture (3C) assay protocol was performed in this study, as described previously (21). 10⁷ cells were washed twice in cold PBS. Formaldehyde was added to cells to a final concentration of 1% and incubated for 10 min at room temperature. Glycine (final concentration 125 mM) was added to stop the cross-linking reaction. Nuclei were collected from the cross-linked cells, and the DNA was digested overnight at 37°C with BglII and subsequently heat inactivated for 20 min at 65°C. Samples were diluted 1:40 into ligation buffer and ligated overnight with T4 DNA ligase (20,000 New England BioLabs units) at 16°C. Proteinase K (final concentration to 200 µg/ml) was added to the ligation reactions, and the DNA/protein complexes cross-links were reversed by incubating overnight at 65°C. The DNA was purified by phenol/chloroform extraction and ethanol precipitation. Quantitation of the 3C products was performed by real-time PCR using a five-point standard curve (21). Standard curve templates for the 3C products were generated in vitro by restriction enzyme cleaving and ligation of A20 genomic DNA (21, 22). All primer combinations (Supplemental Table I) were tested prior to use in the 3C assay to determine whether they could efficiently amplify a single product from the randomly ligated A20 genomic DNA fragments. Data were normalized as cross-linked frequency compared with the input DNA of the standard curve.

To identify mouse B cell SEs, H3K27ac data from CD43⁻ B cells were reanalyzed (34). Enriched H3K27ac peaks for each replicate were determine using MACS2 v 2.1.1 (35), and all within 12,500 bp were merged using the bedtools v2.24 merge function (36). Merged peaks were filtered against the ENCODE mm9 blacklisted regions (37) and the read depth at each peak annotated using the GenomicRanges v1.34 (38) and rtracklayer v1.24 (39) Bioconductor packages in R v3.5.2. Read depths were averaged for each cell type, normalized to the maximal peak depth, and ranked. The slope was calculated for the normalized ranked enhancer list, and enhancers with a slope of 1 or greater were classified as an SE as previously defined (40). The genome plot figure showing the H2-Aa and H2-Eb1 locus was generated with code that is available at https://github.com/cdschar using previously published datasets, as referenced in the Fig. 1 legend.

Previous analysis of the murine H2-Aa and H2-Eb1 genes identified two strong CTCF sites that flanked the genes and one moderate site (termed C15) located between them (24). The three sites interacted with each other and with the H2-Aa, H2-Ab, and H2-Eb1 promoters (24). To explore the region further, bioinformatic analysis of published histone ChIP-sequencing (ChIP-seq) data (34, 41) revealed that the murine MHC-II intervening sequences had high levels of the activation-associated histone mark H3K27ac in B cells that were lost in plasma cells (Fig. 1A). The two flanking broad histone H3K27ac peaks were associated with proximal promoter regions of H2-Aa and H2-Eb1 (labeled in this study as regions a and d, respectively), and one of the two intervening regions was associated with C15 (region c). The remaining H3K27ac set of peaks in the intervening region spanned ∼15 kb of the sequence (region b). H3K4me1 modifications (42), which are associated with enhancers and promoters, were found extensively across the entire region, including all three CTCF peaks and within the body of the genes (Fig. 1A). H3K4me3 is associated with active promoters and was found in B cells at the promoters and within several locations within the intervening I-A/E region. Correlating with the loss of MHC-II expression, all H3K4me3 peaks were greatly decreased in plasma cells. Chromatin accessibility as determined by assay for transposase-accessible chromatin using sequencing (43) showed regions of high accessibility within all of the regions described in this study, including C15, the H2-Aa, and H2-Eb1 promoters, and flanking CTCF sites in B cells. Reduced accessibility across the locus was observed in plasma cells.

The high levels and breadth of active histone modifications suggested that parts of the region could be considered an SE. Analysis of all B cell H3K27ac peaks identified 506 SEs and 12,254 conventional enhancers (Fig. 1B). Of the SEs, the region associated with H2-Aa scored the highest in the genome with another MHC-II gene, H2-Ab1, the fifth highest ranked locus. As noted above and in Fig. 1A, we have divided the H2-Aa/Eb1 SE into four parts, separating the gene promoters from C15 and the novel region associated with region b. In this study, the entire region will be referred to as the IA/IE-SE.

Because of the intriguing levels of histone ac and the fact that C15 was homologous to the human XL9 MHC-II insulator, we sought to explore its role in regulating MHC-II gene expression by creating a genetic knockout of the region using homologous recombination technologies in C57BL/6 mice. Founder lines were crossed onto a Cre recombinase transgenic strain driven by the EIIA promoter to delete C15 in all cells of the offspring. Two founder strains of mice resulted in which homozygous deletions were verified by Southern blotting and PCR (Supplemental Fig. 1). The lines were termed ΔC15 A and B. Both mouse lines produced identical results in early experiments, and therefore, the homozygous mouse strain ΔC15 A was characterized further and will be referred to in this study as ΔC15.

The effects of homozygous C15 deletion were first assessed on the chromatin structure of the region by traditional ChIP assays using the chromatin from splenic B cells of C57BL/6 WT and ΔC15 A mutant mice. A total of eight loci/sites covering the promoters of H2-Aa (site 2) and H2-Eb1 (site 6) and the intervening regions were examined (Fig. 2). Additionally, a negative control site was chosen from the mouse Hox locus. As expected from the ChIP-seq data, high levels of H3K27ac were observed at ChIP sites 2, N1 (a putative PU.1 binding site within region b of the IA/IE-SE analyzed below), C15, and 6 in WT B cells. In contrast, a signification reduction was observed in ΔC15 mice at each of these sites. High H3K4me1 levels were detected at ChIP sites N1, C15, and 5 in WT cells. A significant decrease in H3K4me1 was observed in ΔC15 B cells compared with WT. H3K4me3 levels were significantly diminished but still substantial over the promoters in ΔC15 cells. Last, although CIITA binding was significantly reduced at the H2-Aa promoter region, there was no significant difference at the H2-Eb1 promoter. Thus, active chromatin modifications of these sites (MHC-II promoters) and site N1 was largely dependent on the presence of C15 in the locus.

Depletion of CTCF from human Raji B cells resulted in reduced expression of all MHC-II genes but did not affect the expression of CIITA or RFX5, two key modulators of MHC-II expression (21). Exactly similar results were found in mouse B cells (24). Thus, it was hypothesized that the ΔC15 mice would have defects in MHC-II expression. Cells were prepared from the bone marrows of C57BL/6 and ΔC15 mice that represented the B cell developmental spectrum (CLP, pre-pro, pro-B, pre-B, immature, and mature B cells), and surface MHC-II expression was examined by flow cytometry (Fig. 3). The results displayed no significant differences between the WT and ΔC15 strains for any of the B cell developmental subsets. The same was observed in the mature B cell subsets of splenic marginal zone B and follicular B cells. Because defects in MHC-II expression could affect developing T cells in the thymus, developing thymic T cells in ΔC15 mice were examined and compared with WT animals. The results shown in Supplemental Fig 2A indicated that T cell development was normal, with no changes in the percentage of CD4 and CD8 single or double positive T cells. Thus, whereas depletion of CTCF protein from cells disrupts the chromatin interactions, leading to a reduction in MHC-II expression, deletion of a single CTCF site C15 did not change the expression of MHC-II genes in B cells or developing B cells or result in changes in developing thymic T cell populations.

Because CTCF–CTCF interactions, including those involving C15, are disrupted during B cell differentiation to plasma cells, the effect of C15 deletion on this process was also evaluated. Splenic B cells from WT and ΔC15 mice were isolated and examined for MHC-II expression and subjected to ex vivo differentiation in the presence of LPS, IL-4, and IL-5. Splenic B cells showed no differences in surface MHC-II expression (Fig. 4A). Ex vivo differentiation resulted in ∼30% CD138⁺B220^mid/lo plasmablasts, forming with no differences between WT and ΔC15 cells (Fig. 4B, 4C). On day 3 of culture, MHC-II expression was significantly and equally reduced in WT and ΔC15 as expected (Fig. 4D, 4E). Last, no differences in H2-Aa, H2-Eb1, or CIITA mRNA levels were observed in these cultures when comparing WT to ΔC15 (Fig. 4F). Thus, the deletion of C15 does not affect the development of B or T lymphocytes or expression of MHC-II on B cells or alter the course of B cell to plasmablast differentiation.

Because MHC-II gene products can be induced on many cell types following treatment with IFN-γ (25, 44), bone marrow–derived macrophages were first examined. Bone marrow macrophages were cultured ex vivo with IFN-γ over a 3-d time course. Not surprisingly, MHC-II surface expression was induced in cells derived from C57BL/6 WT mice (Supplemental Fig. 2B, 2C). Similarly,bone marrow–derived macrophages from ΔC15 mice also induced MHC-II genes with no changes in the kinetics or amount of product. The levels of H2-Aa, H2-Eb1, and CIITA mRNA were also similar during the induction time course between WT and ΔC15 macrophages (Supplemental Fig. 2D). These data suggest that the ΔC15 does not alter MHC-II induction by IFN-γ in bone marrow–derived macrophages.

Nonimmune cells can also be induced by IFN-γ (13, 44). As such, cells may undergo a different developmental pathway, and the effect of C15 deletion may be distinct. Therefore, MEF were generated from day-12 embryos of WT and ΔC15 mice. Cells were incubated in a 3-d time course and analyzed by flow cytometry for surface MHC-II and mRNA (Fig. 5). Cells derived from WT embryos showed a peak of MHC-II surface expression at 24 h and mRNA expression for H2-Aa, H2-Eb1, and CIITA that was stable from 24–72 h (Fig. 5). However, the induction of MHC-II protein or mRNA was remarkably absent in ΔC15 MEFs. Because of the unexpected result, this experimental protocol was repeated four times over the course of several generations of ΔC15 mice and yielded the same result. Taken together, these data suggest that the developmental lineage that leads to embryonic fibroblasts is distinct from the path that yields the immune cells tested and is dependent on C15 for MHC-II expression.

To further understand the phenotype observed in MEFs, ChIP assays assessing the levels of H3K27ac, H3K4me1, and H3K4me3 were performed across the IA-IE-SE following exposure of MEF cells to IFN-γ. In response to IFN-γ, WT MEFs increased levels of H3K27ac and H3K4me3 at H2-Aa and H2-Eb1 promoter regions. This response was absent in MEFs derived from ΔC15 mice (Fig. 6). Additionally, in WT cells, site N1, which is located in the region b of the SE, displayed high levels of H3K27ac and H3K4me1 in the absence of IFN-γ. N1 also showed an increase in both of these marks in response to IFN-γ. In ΔC15 MEFs, N1 had significant levels of H3K27ac and H3K4me1, but there was no induction upon IFN-γ treatment. The HoxA9 locus was used as a locus control for these assays, and no difference in activity was observed. These results point to a dysfunction in the ability to induce MHC-II genes in the MEFs because of a C15 mutation but also support a role for the region defined by N1 as an active region in B cells and in response to IFN-γ in MEFs.

In our previous work examining both the human and murine MHC-II loci (5, 21, 22, 24), CTCF-bound elements interacted in a distance-dependent manner, forming long-range chromatin loops across the MHC-II locus. The occurrence of multiple potential CTCF binding sites suggested the possibility that the loss of one CTCF site might be compensated by another and could explain the reasons for MHC-II expression not being altered by the ΔC15 mutation. Thus, to determine whether changes in levels of interaction between the CTCF-bound elements, promoters, or any other interacting sites occurred in the absence of C15, a quantitative 3C assay developed previously for this system (21, 22) was employed on splenic B cells of the ΔC15 mouse and compared with the WT. In this study, promoter region restriction fragments were used as “anchors” to determine the regions with which they interacted. CTCF binding sites (C12, C13, C14, C15, C16, and C17) described previously (24) and site N1, which showed high levels of H3K27ac and contains a consensus PU.1 binding site, were queried as potential 3C interaction targets across the locus (Fig. 7). Self-ligated anchor fragments were used to display restriction digestion/ligation efficiency between cross-linked and non–cross-linked samples. The results showed that the H2-Aa promoter interacted with sites C14, N1, C15, and C16 in WT and ΔC15 B cells to roughly the same degree. The H2-Aa promoter region appeared to interact more strongly with C16 in ΔC15 cells, but this was not statistically significant. Similarly, with the exception of site N1, H2-Eb1 interacted with the CTCF sites irrespective of whether the source of B cells was WT or ΔC15. In this study, site N1 showed increased interactions with the H2-Eb1 promoter in the absence of C15 compared with WT cells.

While generating the ΔC15 mouse, CRISPR–Cas9 was used to make a similar mutation in the B cell lymphoma line A20, a cell line that expresses high levels of MHC-II transcripts and was used previously to study murine MHC-II expression (24). Two independent A20 mutant cell lines were created. Like the ΔC15 mouse B cells, A20-ΔC15 cells showed no differences in I-A expression surface or mRNA levels from WT A20 cells (Supplemental Fig. 3A, 3B). 3C assays were performed using the two A20-ΔC15 lines as above (Supplemental Fig. 3C, 3D). The results revealed that in WT B cells, the H2-Aa promoter region highly interacted with C14, N1, and C15 but not with C16. In both mutant A20-ΔC15 lines, the H2-Aa promoter retained its interaction with C14 and N1, but now, it interacted with C16, indicating that there was a reconfiguration of the interactions to a more distal CTCF site. H2-Eb1 interacted with C15, C16, and with N1 in WT A20 cells. In the A20-ΔC15 cells, H2-Eb1 interactions with C14 and N1 were increased in the absence of C15. Together, these data revealed that deletion of C15 changes the looping structures in the locus and suggests that CTCF sites may function in a redundant manner when one is missing.

To further examine the role of C15 with respect to chromatin architecture in the inducible MEF system, 3C assays were performed on MEF cells derived from WT and ΔC15 mice before and after IFN-γ induction (Fig. 8). The promoters of H2-Aa and H2-Eb1 were again used as anchors in 3C assays (Fig. 7A). In the absence of IFN-γ, the H2-Aa promoter was able to interact with the sites C14, N1, C15, C16, and C17 to various levels in WT MEF cells. Upon IFN-γ treatment, the interaction between H2-Aa and N1 was nearly doubled. Similar results were also observed with interactions between H2-Eb1 promoter in response to IFN-γ. Whereas most interactions between the promoters and the other elements remained unchanged in ΔC15 cells, interactions with N1 were no longer increased upon IFN-γ. Thus, like in B cells, C15 plays a role in organizing the local CTCF chromatin interactions, and its loss results in compensated changes with neighboring interacting sites. However, unlike in B cells, the change in interactions was not sufficient to rescue the IFN-γ–mediated MHC-II expression defect in ΔC15 MEFs.

Bioinformatic analysis of existing B cell ChIP-seq datasets showed that the N1 site is substantially enriched for the binding of the transcription factor PU.1 (encoded by Spi1) (Fig. 9A). PU.1 is expressed in a hematopoietic lineage–specific manner, and several studies showed that PU.1 is positively involved in MHC-II regulation through the transcription of CIITA in dendritic cells, B cells, mast cells, and activated T cells (45–50). To validate this finding, traditional PU.1 ChIP assays were performed using chromatin prepared from A20 and P3X cells (MHC-II–expressing B cells and MHC-II–negative plasma cells, respectively). The results (Fig. 9B) indicated that the N1 site is highly enriched for PU.1 binding in MHC-II–positive B cells but not MHC-II–negative P3X cells, in which only a background level of Ab binding was detected. These data suggest a role for PU.1 and N1 in regulating MHC-II expression.

To further examine the relationship between the N1 region and C15, the N1 region was deleted in A20 cells using CRISPR/Cas9 technology, creating two independent cell lines (A20-ΔN1-1 and A20-ΔN1-2). To show that PU.1 binding was dependent on N1, A20-ΔN1 cells were used in a ChIP assay to determine PU.1 occupancy levels. As observed, both A20-ΔN1 cell lines were not occupied by PU.1 at the N1 site, and no changes in PU.1 occupancy were observed at the other sites tested (Fig. 9C). 3C assays were again employed to assess the regional chromatin interactions. The results showed that in WT A20 cells, the H2-Aa promoter interacted with sites N1, C14, and C15, and H2-EB1 interacted with C14, N1, and C16 (Fig. 9D). In A20-ΔN1 cells, increased interactions were reproducibly observed between H2-Eb1 promoter and C14. Thus, the loss of N1 also altered the chromatin configuration of the locus.

To determine the surface expression of the MHC-II level in A20-ΔN1 cells, A20-ΔN1-1 and A20-ΔN1-2 cells were examined by flow cytometry. Compared with WT A20 cells, both A20-ΔN1 cell lines showed reduced levels of MHC-II gene expression (Fig. 9E). Real-time RT-PCR using RNA isolated from WT A20 and A20-ΔN1 cells also showed reduced expression for H2-Aa and H2-Eb1 but not CIITA in the mutants compared with the WT A20 cells (Fig. 9F). These data demonstrate a role for the N1 region of the IA/IE-SE in regulating the transcription of MHC-II genes.

Common regulatory mechanisms have defined the coordinate regulation of MHC-II genes in both humans and mice, including the promoter proximal cis-regulatory elements and their transcription factors, as well as a role for CTCF (reviewed in Refs. 5 and 20–23). Recently, an SE located in the intergenic region between HLA-DRB1 and HLA-DQA1 of the human MHC-II locus was found to play a role in regulating MHC-II expression (28), thereby introducing a novel aspect of regulatory control to this gene system. Bioinformatic mining of histone H3K27ac ChIP data revealed that MHC-II sequences encompassing the H2-Eb1 and H2-Aa genes was a top-scoring SE in mouse B cells and suggested that it may play an analogous function in the mouse. Experiments in this study were therefore focused on determining the role of the intergenic sequences (region b containing site N1 and C15 CTCF site) in regulating murine MHC-II gene expression.

To investigate C15, we created several model tools, including a genetic knockout mouse and CRISPR/Cas9-modified cell lines. The results using both models showed that the C15 deletion did not affect the numbers of B and T cells in the developing mouse or the expression of MHC-II genes on B cells or macrophages. Even though the loss of C15 did not alter the overall transcription level of the neighboring MHC-II genes in these cells, it did have an effect on the local chromatin structure by reducing the active histone marks associated with the promoter regions and the IA/IE-SE. CIITA binding at the promoters was also reduced but still present. This indicates that C15 sequences serve to maintain the steady level of the active chromatin state in its surrounding regions. 3C analyses showed that the likely reason for this surprising observation was that the MHC-II CTCF sites played redundant roles. Thus, with a plethora of CTCF sites in the MHC locus, one CTCF site could functionally replace a deletion of a neighboring site. These results are similar to those observed in human B cell lines in which XL9 was deleted by CRISPR/Cas9 methods (28), resulting in changes in active histone marks but not in alterations of HLA-DR or HLA-DQ expression. These data suggest that the ΔC15 and XL9 are functionally equivalent and that the additional CTCF sites in both humans and mice can functionally compensate for one another.

In addition to their constitutive expression in B cells, MHC-II genes are expressed in APCs, such as macrophages, and can be upregulated in macrophages and induced in nonimmune cells by IFN-γ. Thus, we hypothesized that although the B cell lineage might be unchanged with respect to MHC-II expression in the ΔC15 mutant, this may not be the case for other cell types and those induced by IFN-γ. However, MHC-II expression in macrophages at baseline or in response to IFN-γ treatment was not altered. Instead, IFN-γ treatment of embryonic fibroblasts derived from ΔC15 mice failed to upregulate MHC-II genes. This failure to induce MHC-II genes was not due to an IFN-γ signal transduction mechanism as CIITA expression was induced to WT levels. Clues to the failure were provided by observing losses of histone marks associated with gene activation. Additionally, unlike in WT cells, in which IFN-γ treatment resulted in increased 3C interactions between N1 and the MHC-II promoters, such increases did not occur in the ΔC15 mutant cells. Thus, immune cells that would normally express or upregulate MHC-II in response to IFN-γ were not affected by deletion of C15. In contrast, the failure to induce MHC-II induction in MEFs by IFN-γ suggests that the formation of the chromatin structure during the early stages of development from which MEFs are generated was compromised. The difference observed between MEFs and macrophages may relate to embryonic lineages from which each of these tissues is derived.

The 3C analysis did, however, reveal a role for region b of the IA/IE-SE. As with the human locus, region b of the IA/IE-SE was critical for maximal MHC-II expression and formed interactions with the neighboring CTCF sites and MHC-II promoters. The data therefore suggest that the MHC-II SE is required for maximal expression and has evolved as a critical regulatory component with the locus. Like other SEs, the IA/IE-SE is a complex element with potentially many transcriptional regulatory functions that can be hypothesized by its four peaks of H3K27ac enrichment. Fortuitously, using available databases, we identified the PU.1 binding site N1 as a potential core component within region b of the IA/IE-SE. Indeed, ChIP assays showed that PU.1 bound to this putative site. PU.1 was previously shown to be indirectly important for MHC-II expression as it regulates CIITA expression in B cells by binding to promoters I and III of the CIITA gene, depending on the cell type (45, 46). Thus, PU.1 is likely playing a dual role by regulating in a feed-forward circuit both the expression of CIITA as well as MHC-II expression. To further investigate the role of region b and the N1 site, we produced an N1 deletion mutant from the MHC-II–positive B cells (A20) using CRISPR/Cas9 technology. The loss of N1 resulted in reduced levels of MHC-II expression and alterations in the interactions between the promoter regions and the local CTCF sites. Thus, like the DR/DQ-SE of the human locus (28), the N1 region of the IA/IE-SE interacts with the local MHC-II promoters and is required for maximal MHC-II expression.

Together with the data generated for the human XL9 and DR/DQ-SE (28), the results in this study point to a complex organization across the MHC-II locus that is required for expression of these important immune system genes. CTCF sites are located throughout the locus and serve as key points of regulation and chromatin organization within the MHC. CTCF sites interact with each other and recruit the cohesin complex, which binds to CTCF and encircles DNA strands to stabilize a loop and long-distance interactions. We previously observed that cohesin binds to the human MHC-II CTCF sites, and knockdown of cohesin subunits resulted in lower MHC-II gene expression (23). Cohesin also has an ATPase-dependent function that will extrude DNA on one side of the loop, moving enhancers toward their target promoters (reviewed in Ref. 18). Thus, one can speculate that interactions between the MHC-II CTCF sites set up the genes to be expressed with the cohesin complex, serving to bring the enhancers toward the promoters to enhance transcription of the locus. In the murine locus, C15 may serve that purpose when present, but when absent, other CTCF sites fill that void. It is intriguing that disruption of CTCF binding to C15 in plasma cells is also accompanied by a complete rearrangement of all MHC-II locus–specific CTCF–CTCF interactions. Such rearrangement would potentially alter the ability of the CTCF/cohesin complex system from enhancing the expression of MHC-II genes.

In summary, the work presented in this study identified and provided an additional layer of regulation to murine MHC-II gene expression and substantiates a model predicted by human studies in which an SE is critical to the regulation of these genes and is maintained during the divergent evolution between humans and mice. We also provide evidence that the multiple CTCF sites can function in a redundant manner. Such a property may have allowed multiple MHC-II gene pairs to be gained or lost during evolution as the ability to share CTCF sites would allow gene duplication events to occur without the need to replicate long-distance interactions and control mechanisms.

We thank all laboratory members for critical and sincere comments during this study, in particular Tian Mi for help with sequence analysis.

The authors have no financial conflicts of interest.